Aug 9 – Sep 8, 2026 · 5 items
The AI week, for a management consultant
A real edition, written for: Management Consultant
The one thing
OpenAI's new model now operates browsers and spreadsheets directly, which changes how much of your research and deck prep you can hand over.
OpenAI's new model works inside spreadsheets and browsers, not just chat
OpenAI released GPT-6 Astra, a frontier model built to operate computers directly. It works across browsers, spreadsheets, websites and desktop applications. It can fill forms, update CRM records, run web research and produce documents. OpenAI says this reduces the need for hand-built connectors to each business system. President Greg Brockman told a press briefing that the company is now in the AGI era. Rollout starts Thursday for enterprise customers in the Daybreak gated program. Paid ChatGPT tiers, the OpenAI API, AWS Bedrock and Microsoft Azure follow in coming days. Brockman argued buyers should compare price per completed task rather than per token. OpenAI omitted GDPval, its own benchmark for real-world occupational work. OpenAI also paused some frontier training for about two weeks after the Hugging Face incident and tightened infrastructure controls.
Why it matters for you
You already use ChatGPT for research and deck prep. This model does the clicking too. It can pull web research and produce documents rather than hand you text to paste.
- Access starts with gated enterprise customers, then paid ChatGPT tiers over the following days
- Your firm's IT will likely decide when Business or Enterprise seats get it, so check before promising a client
- OpenAI skipped its own real-work benchmark, so treat the occupational claims as unproven
Try this
Take one industry primer you built by hand last month and rebuild it in ChatGPT to compare quality.
Paste this into your AI tool
Build me a 6-slide industry primer on [industry], for a client operating model review. For each slide: a headline finding, three supporting facts with the source named, and one implication for how a company in this sector should be organised. Flag anything you are unsure about instead of smoothing it over. Output as plain text I can move into slides.
Meta cancelled the layoffs that assumed AI agents would do the work
Meta cancelled the second wave of layoffs planned under an internal reorganisation that would have shifted much of the daily work of thousands of employees onto AI systems overseen by small expert teams. The company still cut about a tenth of its staff in May, but internal measures showed autonomous AI agents were not delivering the expected productivity, while technical and security incidents rose. Zuckerberg told staff in July that the pace of AI agent progress had been misjudged.
Why it matters for you
This is the case study your transformation clients will ask about. A large company bet the operating model on agents and pulled back. Incidents and fix times went the wrong way.
- Code output rose sharply while defects and recovery time rose faster, which is the trap in agent-led redesign
- Useful counterweight when a client board wants headcount savings written into the business case
- The pullback came after the first cut had already landed, so the sequencing lesson is real
Try this
Add an agent-productivity reality check to your standard operating model diagnostic questions.
Paste this into your AI tool
I advise clients on operating model redesign. Draft ten diagnostic questions I should ask before a client assumes AI agents will absorb routine work. Base them on failure modes like rising incident rates, longer recovery times, quality debt and falling staff sentiment. For each question, say in one line what a bad answer sounds like.
Sources
Only a fifth of large firms have scaled AI beyond one business unit
Gartner found that only 22% of large organisations have scaled AI across several business units. It surveyed more than 1,300 leaders at firms with over $50 million in revenue, between January and April. Spending plans are undented, with 85% of technology leaders raising AI budgets next year. About 11% of respondents could not say what they spent on AI in 2025. Gartner's Tina Nunno warned that weak measurement tied to business outcomes wastes resources. Firms that track returns continuously and shut down weak projects reported gains on 81% of initiatives. Popular uses such as cybersecurity, threat detection and IT service desk automation often return less. The best returns came from IT asset and cost optimisation, synthetic data generation, and automated code generation. Separate reports from Infosys and Deloitte found similar gaps in measurement and readiness.
Why it matters for you
Your clients are mostly in the stuck majority. The survey names what separates the ones getting returns: continuous measurement and killing weak projects.
- A tenth of leaders could not say what they spent last year, which is a governance finding you can use directly
- Popular use cases like service desk automation returned less than cost and code optimisation
- Budgets are still rising regardless, so the advisory opening is discipline, not enthusiasm
Try this
Turn the portfolio-discipline finding into a one-page maturity check you can run in a client kickoff.
Paste this into your AI tool
Draft a one-page AI portfolio discipline assessment I can run in a client kickoff workshop. Five dimensions: spend visibility, outcome measurement, project stop criteria, ownership, and scaling beyond one business unit. For each, give three maturity levels described in plain language a COO would recognise. Keep it to one page.
Nvidia is buying Hugging Face
NVIDIA has agreed to acquire Hugging Face, the open model sharing platform. Chief executive Jensen Huang announced the deal on NVIDIA's blog. He put the price at just under $13 billion. NVIDIA says Hugging Face will stay open to the whole AI industry. Developers will keep their choice of models, frameworks, clouds and chips. NVIDIA compute will not be required to build or deploy through the platform. Hugging Face will continue to support open source and open weight models from all builders. NVIDIA says its infrastructure and engineering can improve reliability, safety, model evaluation and deployment. The Hugging Face team and its brand will remain. No closing date or regulatory timeline was given.
Why it matters for you
Where open models live is now owned by the chip supplier. Clients who chose open weights to stay independent will want your read on that.
- The deal is not expected to close until the first half of 2027, subject to approval
- Nvidia says compute choice stays open, but analysts advise watching for tighter tooling ties later
- Worth a line in any vendor-concentration section of a technology strategy paper
Try this
Draft the two-paragraph client note on what this changes for open-model strategies.
Paste this into your AI tool
Write a two-paragraph client note for a COO on Nvidia's agreement to acquire Hugging Face. Cover: why an organisation might have chosen open-weight models in the first place, and what supplier concentration risk they should now watch for. No hype, no speculation about the deal closing. Plain business language.
Sources
EU AI Act transparency duties are now enforceable, and the AI Office can ask
The European Union's AI Act moved into enforcement on 2 August 2026, with disclosure duties now applying to chatbots and to content produced by AI. The EU's AI Office can request information from covered companies and ask for access to their models, though it has not yet pursued anyone for misconduct. Anthropic, Google, Meta, OpenAI and Microsoft have each described compliance steps, including watermarking generated text. Rules for high-risk uses such as education, biometrics and migration arrive only in December 2027 and August 2028.
Why it matters for you
You advise from Switzerland, so your clients selling into the EU are in scope even if your firm is not. Disclosure duties on chatbots and AI content now bite.
- The AI Office can formally request information and model access, though no case has been brought yet
- High-risk rules for areas like biometrics and education land much later, in 2027 and 2028
- A transformation programme designing customer-facing chatbots needs the disclosure line in scope now
Try this
Add an AI Act disclosure checkpoint to the scoping template for any client-facing chatbot workstream.
Paste this into your AI tool
I advise clients based outside the EU who sell into it. Draft a short scoping checklist covering EU AI Act transparency and disclosure duties for customer-facing chatbots and AI-generated content. Six to eight items, each a question a project lead can answer yes or no. Note which items apply now and which relate to later high-risk phases.
Sources
Build this
Every week, one small thing to build with AI in something you actually care about. No work in it. Five minutes to set up, and worth keeping if it earns a second run.
A three-day autumn walking route built only from places you have already been, reordered so they connect properly instead of repeating.
Five minutes to set up
Build a three-day walking itinerary in Switzerland using only these places I have already visited: [list five or six valleys, towns or huts you know]. Do not add any location I have not named. Sequence them into a route that works with public transport, give rough walking times per day, and say what makes each day different from the last. If two of my places do not connect sensibly, say so and drop one rather than inventing a link.
- Write down five or six Swiss spots you have actually walked, not ones you mean to visit
- Run it in ChatGPT, and reject the answer if it slips in a place you did not name
- Check the transport connections and walking times yourself before booking anything
- When a day feels wrong, swap that single location and re-run rather than starting over
Yours arrives Thursday.
This one was written for a management consultant. Tell us what you do and the next one is written for you — same news, your job, once a week.